Hardware
"National investment in compute capacity is a new economic imperative... countries are awakening to the need to invest in sovereign AI infrastructure" says Jensen Huang
MicroCloud can scale from three servers to around 50-node clusters and it is lightweight enough, claims Canonical, to run on a developer laptop.
Big claims by Satya Nadella, big news for the industry, but no benchmarks or hard specifications yet.
An eight-way HGX H200 provides over 32 petaflops of FP8 deep learning compute and 1.1TB of aggregate high-bandwidth memory
"Cache misses have this weird massive non-linear effect into how much work the GPUs are doing, because we suddenly need to start recomputing all this stuff."
"People often think theatre is very scary and confrontational. It can be at times. But it's not the way we do it and I think that's why it’s a very unique experience..."
NVIDIA A100 80GB powered GPU instances are “immediately available" and a range more coming very soon.
"SoftBank, you know, is not known as a poster child for good corporate governance"
The chipmaker says it can double the speed of inference on its H100 GPUs